What is AI liability? AI liability is the legal responsibility a company carries when an artificial intelligence system it deploys produces false, defamatory, or inaccurate output that causes real harm. On May 28, 2026, the Regional Court of Munich gave a blunt answer to a question every commercial real estate firm using AI should be asking: when an AI tool gets it wrong, who pays? The court ruled that Google is directly liable for false statements generated by its AI Overviews, treating the machine generated summary as Google's own speech rather than neutral search results. For CRE investors who increasingly lean on AI to underwrite deals, write marketing copy, and screen tenants, this decision turns AI liability from a theoretical worry into a board level risk. For the bigger picture on adopting these tools safely, see our guide to the best AI tools for commercial real estate.
Key Takeaways
- The Regional Court of Munich ruled on May 28, 2026 that Google is directly liable for false claims its AI Overviews generate, classifying the AI output as Google's own content.
- The decision rejects the traditional search engine liability shield and signals that telling users to double check the AI does not transfer responsibility back to them.
- For CRE investors, AI liability exposure is highest in underwriting math, property marketing claims, tenant screening, and legal document drafting.
- AI hallucination rates on CRE financial data remain material, so unverified output in an offering memorandum or investor report can create real legal and reputational damage.
- Human review, enterprise grade tools, audit trails, and tighter vendor contracts are the core defenses that limit AI liability for real estate firms in 2026.
What the German Google AI Liability Ruling Actually Said
The Regional Court of Munich, in case 26 O 869/26, barred Google from spreading false claims about two Munich based publishers through its AI Overviews. The AI had falsely tied the publishers to scams and subscription traps, mixing in details from unrelated companies and inventing connections that appeared in none of the linked sources. The court classified Google as a direct infringer because the AI Overview is its own content, not a list of third party links. As the court framed it, Google built the AI, Google offered it to users, so Google owns what it produces. The judges explicitly found that older German precedent granting search engines limited, indirect liability does not apply to generative summaries, and that a disclaimer telling users to verify results is not enough to escape responsibility. This is a preliminary injunction rather than a final precedent, and Google is appealing, but it is one of the first decisions worldwide to assign direct responsibility for an AI hallucination to the company that built the model. An analysis cited by the New York Times found Google AI Overviews, running on Gemini, are accurate roughly 91 percent of the time, which sounds reassuring until you apply the 9 percent error rate across billions of queries.
Why AI Liability Now Matters for CRE Investors
AI liability matters for CRE because the same hallucination problem that defamed two publishers can surface in a rent roll summary, a valuation memo, or a listing description, and the firm that published it usually faces the consequence, not the AI vendor. Commercial real estate has moved AI from experiment to infrastructure. With 92 percent of corporate occupiers having initiated AI programs and the AI in real estate market projected to reach $1.3 trillion by 2030 at a 33.9 percent CAGR (Source: industry market research), the volume of AI generated CRE content is climbing fast. The Munich ruling tells deployers that the legal system increasingly treats AI output as your own speech. If your underwriting model fabricates a comparable sale, or your marketing assistant overstates a property's net operating income, blaming the model is not a defense. JLL has written about how AI is reshaping the sector in its analysis of the future of corporate real estate in the AI age. For the operational side of this risk, see our analysis of whether you can trust AI to underwrite a deal.
Where AI Errors Create the Most CRE Liability Risk
AI liability risk concentrates wherever an unchecked machine output drives a financial, legal, or public facing decision. Four areas carry the highest exposure for real estate firms.
- Underwriting and financial analysis: An AI that miscalculates cap rate, DSCR, or IRR, or invents a comparable sale, can mislead investors and lenders. Our benchmark study on AI hallucination rates for CRE financial data shows error rates are far from zero.
- Property marketing and listings: AI generated descriptions that overstate condition, zoning, or income can trigger misrepresentation claims, the exact category of harm at issue in the Munich case.
- Tenant and borrower screening: Tools that surface false adverse information about an applicant raise both liability and fair housing exposure.
- Legal and lease drafting: AI hallucinated citations have already drawn court sanctions. See our coverage of AI hallucinations and record court sanctions for what happens when fabricated content reaches a filing.
How CRE Firms Can Limit AI Liability in 2026
The reassuring part is that AI liability is largely controllable through governance rather than avoidance. Five practices do most of the work.
- Keep a human in the loop: A licensed professional should verify every AI claim that informs a financial, legal, or marketing decision before it leaves the building.
- Use enterprise grade tools, not consumer chatbots: Enterprise tiers of ChatGPT, Claude, and Gemini offer data governance, zero retention options, and audit logging that consumer accounts do not.
- Build an audit trail: Log which model, prompt, and version produced each output so you can reconstruct a decision if it is ever challenged.
- Tighten vendor contracts: Review indemnification, accuracy disclaimers, and data use terms, because the Munich court signaled that vendor disclaimers may not shield the deployer.
- Train your team: Most AI liability traces back to an unverified copy and paste, not a model defect.
CRE investors looking for hands on AI implementation support can reach out to Avi Hacker, J.D. at The AI Consulting Network to build this kind of governance directly into their underwriting and marketing workflows.
The Bigger Picture for AI Accountability
The Munich decision is preliminary and appealable, but it points to a clear direction of travel: courts and regulators are steadily shifting liability toward the companies that build and deploy AI, and toward the businesses that rely on it. The EU AI Act reaches its next enforcement milestone in August 2026, Fannie Mae has issued new AI governance rules for lenders, and several US states are adding their own requirements. Our explainer on Fannie Mae's AI governance rules shows how this is already reaching real estate finance. Industry researchers including CBRE continue to document how fast AI adoption is outpacing controls, and the uncomfortable backdrop is that only 5 percent of organizations report achieving most of their AI program goals. That means governance maturity is lagging adoption at exactly the moment liability is rising. If you are ready to deploy AI across underwriting and operations without taking on uncontrolled risk, The AI Consulting Network specializes in exactly this.
Frequently Asked Questions
Q: Does the German AI liability ruling apply to US commercial real estate firms?
A: Not directly, because it is a preliminary injunction from a German regional court. However, it reflects a global trend toward holding AI deployers accountable, and US firms should treat it as an early warning rather than a foreign curiosity.
Q: Who is liable if an AI tool makes an error in my underwriting?
A: In most cases the firm that relied on and published the output bears the responsibility, not the AI vendor. Vendor terms of service typically disclaim accuracy, which is why human verification of every material number is essential.
Q: Can a disclaimer protect my firm from AI liability?
A: The Munich court found that telling users to double check AI results was not enough to deny liability. A disclaimer may help, but it does not replace real review, audit trails, and professional sign off on AI assisted work.
Q: How accurate are AI tools on CRE financial data?
A: Accuracy varies widely by model and task. Even leading models hallucinate on financial data at meaningful rates, so any cap rate, NOI, or comparable produced by AI should be verified against source documents before it informs a decision.